AI-Powered Learning Platform

Subject Syllabus

AL802 (B) - High Performance computing

AIML · Semester 8

Course Modules

Learn topic-wise with Hyper AI and revise using mapped PYQs.

General Purpose cache based architecture - performance metric and bench marks Moors Law, pipelining, super clarity, SIMD Memory Hierarchies Multi core processors, Multi threaded processors Vector processors - Design principle, Max performance estimates, programming for vector architecture Basic Optimizations for serial codes: Scalar profiling, common sense optimizations Simple measures and their impacts, role of compilers, C++ optimizations

Balance analysis and light speed estimates Storage order Algorithm classifications and assess optimizations Case studies for data access optimizations Parallel Computers: Shared memory computers, Distributed memory computers, hybrid systems, Network computers

Data and functional parallelism Parallel scalability - laws, metrics, factors, efficiency and load imbalance Shared memory parallel programming with Open MP: Parallel execution, data scoping Work sharing using loops, synchronization Reductions, loop scheduling and Tasking

Program profiling Performance pitfalls Improving the impact of open MP work sharing constructs Determining overheads for short loops Serilisation and false sharing

Message passing Message and point to point communication Collective communication Non blocking point-to-point communication Virtual topologies Efficient MPI Programming: MPI performance tools, communication parameters Impact of synchronizations sterilizations and contentions Reductions in communication overhead